Chapter 6

The Originality Problem

Standing Out When Everyone Shares the Same Tools

“To be yourself in a world constantly trying to make you something else is the greatest accomplishment.”

— Ralph Waldo Emerson

In 2024, two researchers — Anil Doshi of University College London and Oliver Hauser of Exeter — ran an elegant experiment published in Science Advances. They asked hundreds of people to write short stories; some wrote alone, others could draw on AI-generated ideas. The AI-assisted writers produced individually better stories — rated more creative, better written — and the benefit was largest for the least creative writers. Then came the finding that should keep every professional up at night: taken together, the AI-assisted stories were significantly more similar to one another. Individual quality went up. Collective diversity went down.

That is the originality problem in a single experiment. The tool lifts everyone toward the same competent center — which means the center is now free, and crowded.

The Convergence Machine

Every professional in your field woke up with access to the same generator of writing, analysis, strategy, and concepts. When production capabilities are identical, outputs converge — not because anyone is lazy, but because everyone’s raw material now comes from the same distribution. Sameness is the new default. Which makes genuine originality — a perspective that is irreducibly yours — the scarcest commodity in professional life, and the only one the tools cannot mint.

Be precise about what originality is, because it is not novelty. Novelty is cheap; the model produces endless novelty. Originality is a perspective earned through your particular accumulation of experience, reflection, and failure — the thing that lets you see a familiar problem in a way no one else quite does. It is a function of depth. It cannot be prompted, because it is not in the training data. It is in you, or it is nowhere.

When everyone has the same tools, the only genuine differentiator is the irreducibly human perspective you bring to them.

Voice

In professional life, originality expresses itself as voice: the distinctive way you think and communicate that makes your work identifiable without a byline. Voice is built the same way taste is — immersion, production, feedback — which means it is built by writing more than you strictly need to, not less. A professional who lets AI draft everything is not saving their voice for the important moments. They are letting it dissolve into the same competent average as everyone else’s, one delegated paragraph at a time.

The test is simple and slightly uncomfortable. Read your last significant piece of work and ask: does this sound like me, or like a well-executed average? If someone who knows your thinking read it unsigned, would they know it was yours? The Doshi–Hauser experiment suggests what happens to the professionals for whom the honest answer is no — they become interchangeable at exactly the moment interchangeability became free.

The Older Warning Modern AI Just Repeated

None of this is entirely new, which should be reassuring rather than deflating. Long before generative AI, a substantial body of research on brainstorming — running back to work by Marvin Diehl and Wolfgang Stroebe in the late 1980s — found something that embarrassed decades of corporate practice: people generating ideas together in a group consistently produce fewer, less varied ideas than the same people generating ideas alone and pooling them afterward. The mechanisms researchers identified — production blocking, and a social tendency to converge on whatever was said first — are strikingly close cousins of what the Doshi–Hauser experiment found in AI-assisted writing. Shared context, whether a room or a language model, pulls independent minds toward the same output.

The fix that decades of brainstorming research eventually converged on — generate alone first, converge second — is the same fix this book keeps prescribing for AI. Originality is protected by sequence, not by refusing the shared resource. Form your idea in isolation from the group, or from the model, before you let either one shape what you think is possible.

Depth in Practice: The Point-of-View Pipeline

Voice compounds when production is systematic, so give the weekly five hundred words from the practice below a pipeline to live in. Each week’s piece answers one question from your running list — the beliefs colleagues don’t share, the observations that don’t fit the frameworks. No AI in the drafting; the tool is permitted only afterward, as an editor asked one narrow question: tighten this without changing what it says or how it sounds.

Monthly, the strongest piece gets an hour of real revision and goes somewhere public — a LinkedIn essay, an internal memo, a talk proposal. Quarterly, the three best public pieces get read together, looking for the through-line, because the through-line is your emerging perspective becoming visible — to you, and to the people who will eventually hire you, book you, or quote you for it.

Run the pipeline for a year and two things exist that did not: roughly fifty pieces of unassisted thinking — a voice, maintained under load — and a public record of perspective that no colleague with the same tools and no pipeline can imitate, because the only way to have written it is to have thought it. In a converging landscape, that record is the moat.

The Practice

1. Every week, write five hundred words on a topic in your domain, from your own perspective, no AI. Not for publication — for the maintenance of the perspective itself.

2. Answer in writing: what do you believe about the most important problems in your field that most colleagues do not? What have you observed that the standard frameworks miss? Those answers are the seeds of everything unprompted about you.

3. Before sending your next significant piece of work, run the byline test. If it could have come from anyone with the same prompt, put yourself back in until it couldn’t.